CDS Clearing at ICE:
A Unique
Methodology
By Stan Ivanov and Lee Underwood
Credit default swaps play a vital role in the global economy as hedging tools for credit providers
and others with exposure to a corporate or sovereign entity and as a market-based indicator
of an entity’s financial health. Following the financial crisis of 2008, in which uncertainty about
credit risk and counterparty exposures in the over-the-counter market were a major factor,
IntercontinentalExchange introduced CDS clearing in March 2009 with the launch of ICE Trust, now
known as ICE Clear Credit. Today, ICE Clear Credit and ICE Clear Europe, which began clearing
CDS in July 2009, offer clearing on more than 300 index and single name underlying references.
Through Sept. 9, ICE Clear Credit and ICE Clear Europe have cleared $22.4 trillion in gross notional
value resulting in open interest of $1.56 trillion.
F
rom a risk perspective, CDS differ from exchange-traded and other
OTC derivatives in a number of
critical aspects. For example, reliable endof-day pricing for the CDS is not generally available, making it more difficult for
clearinghouses to assess the value of cleared
contracts. CDS also pose “jump-to-default”
risk, meaning the risk of a default that
would yield a very significant financial payment obligation by CDS protection sellers.
Furthermore, the risk presented by CDS is
asymmetrically larger for protection sellers
(short positions) than for protection buyers
(long positions). Finally, CDS instruments
may be especially vulnerable to decreased
market depth and limited liquidity during
distressed economic conditions.
To manage these risks, ICE Clear Credit
and ICE Clear Europe have developed a
unique methodology for CDS clearing. It is
a dynamic model that self-adjusts to market
conditions, reflects the highly asymmetric risk
profiles of CDS instruments, and captures the
specificities of CDS trading behavior while offering robust portfolio efficiencies.
At the heart of the ICE CDS solution
is a daily auction-style price discovery process in which all clearinghouse members
provide end-of-day (EOD) quotes for all
index and single name CDS instruments
in which they have open interest. From
these quotes, the clearinghouse establishes
final EOD prices for mark-to-market and
variation margin calculations, as well as for
computing initial margin requirements and
guaranty fund contributions. All submitted
quotes are subject to the same instrumentspecific bid/offer width to create a coherent
Futures Industry | November 2011 31
Risk Models: ICE
market representation.
To ensure the integrity of the EOD
quote process, the clearinghouse requires
members on random days to execute trades
if the offer price of one clearing member is
lower than the bid price of another. The importance of establishing a market price for
every product cleared cannot be overstated.
Other price-generation approaches, such as
price polling or model-based interpolation,
can be particularly inadequate for providing a true representation of the market. This
is especially true for infrequently-traded
maturities, coupons or reference credits.
The ICE CDS auction-style price discovery
process has proven to be very effective for
obtaining high-quality EOD prices.
Jump-to-Default Risk
While the problem of settlement price discovery is present in other OTC markets,
jump-to-default risk is unique to CDS.
Jump-to-default risk creates a very strong
asymmetry in the risk profile of protectionsold and protection-bought positions. If
a reference entity credit event is expected
E xhibit 1
by the market, the risk is reflected in the
instrument prices, and the daily mark-tomarket process would capture a substantial part of this risk. If the market does not
expect a near-term credit event, however,
the eventual liabilities associated with sold
protection positions can be significant. Collateralization and mutualization levels that
correctly incorporate loss-given-default are
therefore essential for clearing CDS.
For portfolio/concentration risks, large
position requirements, also known as concentration charges, apply to long and short
protection positions that exceed predefined
notional threshold levels. The concentration charge threshold reflects market depth
and liquidity for the specific index family
or reference entity. Positions that exceed selected thresholds are subject to additional,
exponentially increasing, initial margin requirements. The accelerated initial margin
creates the economic incentive to eliminate
large positions. Alternatively, large directional positions can be subject to full collateralization. For example, the protection
seller can create large exposure only if the
ICE CDS Risk Management Framework
Waterfall approach for managing systemic risk
Membership Criteria and Standards
• Ensure each clearing member has sufficient
operational capabilities, risk management
experience and financial resources.
Mark-To-Market Margin Requirement
• Adjusts through a daily debit/credit process
that reflects EOD market prices.
Initial (Risk) Margin Requirement
Intra-Day Risk Monitoring
• Collateralize potential portfolio loss under
distressed market conditions.
• Identify risk requirement erosion and proactively
maintain proper collateralization level.
Guaranty Fund
• Mutualize losses under extreme but plausible
market scenarios. As of June 30, 2011, the
CDS guaranty funds for ICE Clear Credit and
ICE Clear Europe were $3.1 billion and $2.9
billion, respectively.
Limited One-Time Assessment
• Oblige clearing members to contribute a
limited amount of additional default funding.
Source: IntercontinentalExchange
32
Futures Industry | www.futuresindustry.com
seller fully collateralizes its risk. This approach does not require hard position limits
and does not create any detrimental effects
on market depth and liquidity.
In the event of a clearing member default, the clearinghouse may face additional
risk associated with unwinding CDS instruments with relatively low market depth and
activity. The ICE CDS methodology includes explicit provisions for costs associated
with this liquidity risk. For example, a bid/
offer risk requirement, known as a liquidity
charge, appropriately captures liquidation
costs for directional curve portfolios as well
as well-hedged portfolios. The estimated
liquidity charges are based upon historical
observations and the EOD price discovery
process. The liquidity requirements reflect
the expected exposure to wider bid/offer
widths under extreme scenarios of clearing
member default or defaults.
Scenarios and Simulations
An essential function of any risk system,
especially one relied upon by a derivatives
clearinghouse, is capturing portfolio-level
tail risk. This requires a reliable and robust
approach for estimating tail risk at high
quantiles, e.g. 99% and higher, while recognizing the balance between simplicity,
which may lead to conservatively-biased
risk, and efficiency, which may require more
sophisticated risk modeling techniques.
Simplistic initial margin models often lead
to capital inefficiencies and may remove the
incentive for clearing members and market
participants to maintain “flatter” portfolio
risk profiles.
The ICE CDS risk management methodology is a synthesis of two approaches that
undergird a number of risk management
models now in use at other clearinghouses:
scenario-based stress tests and Monte Carlo
simulations. The scenario-based stress approach is the foundation of CME SPAN,
for example, and generally considers a small
set of hypothetical future scenarios. While
the scenario-based stress approach is useful
for portfolios consisting of a relatively small
set of risk factors, its utility for large portfolios composed of nonlinear instruments
such as CDS is limited by the absence of
an efficient portfolio treatment that can
adequately capture hedging and diversification benefits.
The Monte Carlo framework is the foundation for STANS, the risk management
methodology used by The OCC. Advanced
Monte Carlo simulations can readily incorporate realistic distributional assumptions
about the behavior of underlying risk factors, and thus provide reliable and efficient
portfolio risk measures. The main advantage of the Monte Carlo framework is its
ability to generate large numbers of potential market scenarios and to obtain accurate
estimates for the impact of low-probability
events. By contrast, relying on observedonly scenarios, as in historical simulations,
has some disadvantages: Among other
things, it implies that a name that never
defaulted in the past will not default in
the future. Properly implemented Monte
Carlo methods offer better forecasting because they consider scenarios that follow
realistic distributions with specific statistical
behavior—but are not limited to historical
events. The Monte Carlo approach captures low-probability events that have not
happened previously but may occur in the
future. It provides a consistent framework
the stress-test approach, and monitors the
effects of replacing the large set of Monte
Carlo scenarios with a smaller set of stress
scenarios.
Because it can take several days for the
market to absorb the shock of default in an
OTC market such as such as CDS, the ICE
CDS risk methodology has adopted a fiveday risk horizon at a risk quantile of 99%.
This means that margins are set to cover five
days of adverse price/credit spread movements for the portfolio positions with a
confidence level of 99%.
Dynamic Margining
As a dynamic model, the ICE CDS risk
methodology revises margin requirements
as market conditions and credit spreads
change. The approach differs from other
methodologies, where initial margin is established by the clearinghouse as a fixed
metric tail risk associated with protection
selling, the relatively small number of
members, and the need to create strong
incentives for clearing members to actively
participate in the default management
process. As the main source of systemic
risk in the CDS market, large protection
sellers are required to make very significant
contributions to the guaranty fund. Protecting these contributions in the event of
a member default—that is, avoiding the
possibility that the guaranty fund will be
consumed—provides a powerful incentive
to participate in default management and
to expedite the liquidation of the defaulting member’s portfolio.
The size of the guaranty fund is monitored daily through a series of stress scenarios based on observed market realizations
as well as theoretical scenarios derived from
statistical analysis. Guaranty fund moni-
Because the ICE risk management model is self-adjusting,
new initial margin requirements are computed daily, as
CDS spreads and volatility change.
for stress-testing and for sensitivity analysis
in response to deviations from observed behavior, such as correlation structure changes
and market volatility changes.
The ICE CDS risk management methodology can be viewed as a scenario-based
framework that incorporates concepts and
techniques used in advanced Monte Carlo
simulations. The ICE CDS approach features a small number of scenarios that
provide computational transparency and
efficiency, and that reflect the unique characteristics of the CDS market in terms of
credit spread behavior, risk asymmetry, liquidity, quoting conventions, and trading
and pricing practices. Heavy-tailed asymmetric distributions with time-changing
volatilities are used for projecting hypothetical shocks to CDS spreads combined with
credit term structure changes capturing
different “curve” shapes. For internal monitoring and regulatory reporting, the ICE
Monte Carlo framework provides highquantile benchmark Value-at-Risk and
Expected-Shortfall measures extracted from
20,000 scenarios with a full instrument revaluation at each scenario. It features the
same distributional assumptions as those in
dollar amount per contract and often revised following a post hoc review process.
Because the ICE risk management model is
self-adjusting, new initial margin requirements are computed daily, as CDS spreads
and volatility change. The dynamic behavior of the ICE approach improves the risk
forecasting capacity. The ICE CDS risk
framework also recognizes the importance
of initial margin stability for efficient capital
allocation and seeks to mitigate margin rate
fluctuations.
For the guaranty fund size and contribution determinations, ICE aims to achieve
protection against two clearing members
and three additional reference entities simultaneously entering state of default. The
three reference entities are selected based on
the overall exposure of the clearinghouse
to potential extreme losses resulting from
contagion and systemic risk, as well as from
idiosyncratic credit events. The two clearing
members are selected based on the largest
uncollateralized losses under extreme but
plausible scenarios.
Unlike most clearinghouse guaranty
funds, the ICE CDS guaranty funds are
structured to reflect the extreme asym
toring also includes consumption analysis,
which provides information for plausible
scenarios under which available funds will
be exhausted.
Central clearing is a critical step in the
evolution of the credit derivatives market.
With nearly $25 trillion in outstanding
notional cleared since the launch of CDS
clearing in 2009, ICE Clear Credit and ICE
Clear Europe have substantially increased
the stability and transparency of the CDS
market and contributed to the reduction of
systemic risk. Along with the EOD price
discovery process behind mark-to-market
functions, the ICE CDS approach is a major innovation and a substantial development in the modernization of derivatives
and clearinghouse risk management.
..............
Stan Ivanov is chief risk officer, ICE Clear
Credit. Lee Underwood is director, communications, IntercontinentalExchange.
Futures Industry | N ovember 2011 33